Hanshuang Tong, Jun Li, Ning Wu, Ming Gong, Dongmei Zhang, Qi Zhang
{"title":"Ploutos:利用金融大型语言模型实现可解释的股票走势预测","authors":"Hanshuang Tong, Jun Li, Ning Wu, Ming Gong, Dongmei Zhang, Qi Zhang","doi":"arxiv-2403.00782","DOIUrl":null,"url":null,"abstract":"Recent advancements in large language models (LLMs) have opened new pathways\nfor many domains. However, the full potential of LLMs in financial investments\nremains largely untapped. There are two main challenges for typical deep\nlearning-based methods for quantitative finance. First, they struggle to fuse\ntextual and numerical information flexibly for stock movement prediction.\nSecond, traditional methods lack clarity and interpretability, which impedes\ntheir application in scenarios where the justification for predictions is\nessential. To solve the above challenges, we propose Ploutos, a novel financial\nLLM framework that consists of PloutosGen and PloutosGPT. The PloutosGen\ncontains multiple primary experts that can analyze different modal data, such\nas text and numbers, and provide quantitative strategies from different\nperspectives. Then PloutosGPT combines their insights and predictions and\ngenerates interpretable rationales. To generate accurate and faithful\nrationales, the training strategy of PloutosGPT leverage rearview-mirror\nprompting mechanism to guide GPT-4 to generate rationales, and a dynamic token\nweighting mechanism to finetune LLM by increasing key tokens weight. Extensive\nexperiments show our framework outperforms the state-of-the-art methods on both\nprediction accuracy and interpretability.","PeriodicalId":501139,"journal":{"name":"arXiv - QuantFin - Statistical Finance","volume":"13 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-02-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Ploutos: Towards interpretable stock movement prediction with financial large language model\",\"authors\":\"Hanshuang Tong, Jun Li, Ning Wu, Ming Gong, Dongmei Zhang, Qi Zhang\",\"doi\":\"arxiv-2403.00782\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Recent advancements in large language models (LLMs) have opened new pathways\\nfor many domains. However, the full potential of LLMs in financial investments\\nremains largely untapped. There are two main challenges for typical deep\\nlearning-based methods for quantitative finance. First, they struggle to fuse\\ntextual and numerical information flexibly for stock movement prediction.\\nSecond, traditional methods lack clarity and interpretability, which impedes\\ntheir application in scenarios where the justification for predictions is\\nessential. To solve the above challenges, we propose Ploutos, a novel financial\\nLLM framework that consists of PloutosGen and PloutosGPT. The PloutosGen\\ncontains multiple primary experts that can analyze different modal data, such\\nas text and numbers, and provide quantitative strategies from different\\nperspectives. Then PloutosGPT combines their insights and predictions and\\ngenerates interpretable rationales. To generate accurate and faithful\\nrationales, the training strategy of PloutosGPT leverage rearview-mirror\\nprompting mechanism to guide GPT-4 to generate rationales, and a dynamic token\\nweighting mechanism to finetune LLM by increasing key tokens weight. Extensive\\nexperiments show our framework outperforms the state-of-the-art methods on both\\nprediction accuracy and interpretability.\",\"PeriodicalId\":501139,\"journal\":{\"name\":\"arXiv - QuantFin - Statistical Finance\",\"volume\":\"13 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-02-18\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - QuantFin - Statistical Finance\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2403.00782\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - QuantFin - Statistical Finance","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2403.00782","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Ploutos: Towards interpretable stock movement prediction with financial large language model
Recent advancements in large language models (LLMs) have opened new pathways
for many domains. However, the full potential of LLMs in financial investments
remains largely untapped. There are two main challenges for typical deep
learning-based methods for quantitative finance. First, they struggle to fuse
textual and numerical information flexibly for stock movement prediction.
Second, traditional methods lack clarity and interpretability, which impedes
their application in scenarios where the justification for predictions is
essential. To solve the above challenges, we propose Ploutos, a novel financial
LLM framework that consists of PloutosGen and PloutosGPT. The PloutosGen
contains multiple primary experts that can analyze different modal data, such
as text and numbers, and provide quantitative strategies from different
perspectives. Then PloutosGPT combines their insights and predictions and
generates interpretable rationales. To generate accurate and faithful
rationales, the training strategy of PloutosGPT leverage rearview-mirror
prompting mechanism to guide GPT-4 to generate rationales, and a dynamic token
weighting mechanism to finetune LLM by increasing key tokens weight. Extensive
experiments show our framework outperforms the state-of-the-art methods on both
prediction accuracy and interpretability.